One - USF :: School of Natural Sciences and Mathematics

Journal of Volcanology and Geothermal Research 166 (2007) 217 – 232
www.elsevier.com/locate/jvolgeores
Investigation of the groundwater system at Masaya Caldera,
Nicaragua, using transient electromagnetics and numerical simulation
Richard E. MacNeil a , Ward E. Sanford b,⁎, Charles B. Connor a ,
Stewart K. Sandberg a , Mikel Diez a
a
Department of Geology, University of South Florida, 4202 East Fowler Avenue, SCA 528, Tampa, Fl. 33620, USA
b
United States Geological Survey, mail stop 431, Reston Va. 20192, USA
Received 24 January 2007; accepted 31 July 2007
Available online 14 August 2007
Abstract
The distribution of groundwater beneath Masaya Volcano, in Nicaragua, and its surrounding caldera was characterized using the
transient electromagnetic method (TEM). Multiple soundings were conducted at 30 sites. Models of the TEM data consistently
indicate a resistive layer that is underlain by one or more conductive layers. These two layers represent the unsaturated and
saturated zones, respectively, with the boundary between them indicating the water-table elevation. A map of the TEM data shows
that the water table in the caldera is a subdued replica of the topography, with higher elevations beneath the edifice in the southcentral caldera and lower elevations in the eastern caldera, coinciding with the elevation of Laguna de Masaya. These TEM data,
combined with regional hydrologic data, indicate that the caldera in hydrologically isolated from the surrounding region, with as
much as 60 m of difference in elevation of the groundwater table across caldera-bounding faults. The water-table information and
estimates of fluxes of water through the system were used to constrain a numerical simulation of groundwater flow. The simulation
results indicate that basalt flows in the outer parts of the caldera have a relatively high transmissivity, whereas the central edifice
has a substantially lower transmissivity. A layer of relatively high transmissivity must be present at depth within the edifice in order
to deliver the observed flux of water and steam to the active vent. This hydrologic information about the caldera provides a baseline
for assessing the response of this isolated groundwater system to future changes in magmatic activity.
© 2007 Published by Elsevier B.V.
Keywords: caldera; groundwater; electromagnetic methods; geothermal evaluation; hydrology; volcanic structure
1. Introduction
Volcanic eruptions occur when magma exsolves
sufficient volatiles to accelerate flow by rapid volume
expansion, typically in the upper few kilometers of the
crust. As the magma ascends through the crust it may
interact with shallow groundwater in several ways. First,
degassing of volatiles from the magma may create a
⁎ Corresponding author.
E-mail address: [email protected] (W.E. Sanford).
0377-0273/$ - see front matter © 2007 Published by Elsevier B.V.
doi:10.1016/j.jvolgeores.2007.07.016
pressure gradient that can actually drive groundwater to
the surface (Delaney, 1982; Newhall et al., 2001). Sudden
onsets of spring discharge have been observed during the
initial stages of several volcanic eruptions, such as the
eruption of Mt. Pinatubo, Philippines; Usu volcano,
Japan; and the Soufriere Hills volcano, Montserrat
(Newhall et al., 2001; Shibata and Akita, 2001; Sparks,
2003). Second, magma can heat groundwater directly,
sometimes resulting in phreatic eruptions. Such phreatic
eruptions often herald the onset of episodes of volcanic
eruption before magma reaches the surface, as was the
218
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
case at Cerro Negro volcano, Nicaragua, in 1995 (Connor
et al., 1996). Transient boiling of groundwater may
continue for millennia when magma exists in buoyant
equilibrium in the shallow crust, creating shallow
hydrothermal systems (e.g., Hayba and Ingebritsen,
1997; Goff and Janik, 2000; Ingebritsen et al., 2006).
Finally, direct interaction between groundwater and
magma, particularly in confined aquifers, may initiate a
fuel-coolant reaction that results in extremely violent
phreatomagmatic eruptions (Morrissey et al., 2000). Thus,
knowledge of the interaction between groundwater and
shallow magmatic systems can be crucial to understanding
eruptions and the nature of heat and mass transfer in
volcanic systems.
Nevertheless, the distribution and flow of groundwater is poorly known for the vast majority of active
volcanoes (Hurwitz et al., 2003), and hydrologic
properties of active volcanoes are rarely characterized.
This situation persists because such volcanic systems
are not typically developed for their groundwater
resources. The fact that these systems need to be
characterized with scant or nonexistent borehole access
Fig. 1. Outline of Masaya caldera rim with regional groundwater contours. Contours are interpolated across caldera based on water level data in wells
and from a regional ground water flow model developed by the Instituto Nicarguense de Acueductos y Alcantarillados (INECAR) and the Japanese
International Cooperation Agency (JICA). Regional groundwater level varies from ∼190 m above sea level south of the caldera to ~130m above sea
level north of the caldera (ENECAL and JICA; 1993). Inset shows location of Masaya volcano along a portion of the Central American volcanic arc in
Nicaragua within the Nicaragua Depression, a NW-trending fault zone resulting from oblique subduction of the Nazca plate beneath the Caribbean
plate.
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
219
Fig. 2. Topographic map of Masaya caldera with locations of transient electromagnetic (TEM) soundings and profile lines. The caldera rim is inferred
on the northern and southern sides. Inset shows details of the edifice and active crater region. The dashed line indicates the approximate trace of a
fracture system on the NE Flank of Masaya volcano. Lw temperature fumaroles occur in the area of TEM Station 1, along this fracture system.
suggests a need for remote sensing via geophysical
methods such as transient electromagnetic (TEM)
soundings along with innovative inversion techniques.
The interpretation of resistivity in volcanic systems is
complicated because the solutions are commonly nonunique (Kauahikaua, 1993 and Lénat et al., 2000).
Resistivity in volcanic systems is affected by large
temperature gradients, multi-phase flow, hyper-saline
brines, and occurrences of clay-rich alteration minerals.
Separating the effect of these variables in order to
determine depth to the water table can be problematic.
Here, we employ TEM to constrain depth to the
groundwater table at Masaya Caldera, Nicaragua, one of
the largest active basaltic calderas on Earth (Fig. 1). At
Masaya caldera, conditions for interpretation of resistivity
soundings are less problematic than at some volcanic
systems, as there is generally an absence of saline water
and clay alteration in the Holocene stratigraphic section,
except for their probable existence in the hydrothermally
active zone located near the active vent.
We present estimates of depth to the groundwater
table within the caldera based on interpretation of 29
TEM soundings from 30 TEM sites. These data are
compared with water levels measured in wells outside
the caldera. Uncertainty in depth estimates is constrained by using alternative one-dimensional inversion
methods. Our results suggest that Masaya caldera is
isolated hydrologically from the surrounding regional
groundwater flow system by its caldera-bounding faults
and that the caldera's water table is a subdued reflection
of the topography, except near the active Santiago vent,
where dramatic gradients occur.
220
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
2. Masaya Caldera: a volcanic and hydrologic system
Masaya volcano, Nicaragua (11.98°N, 86.15°W) is
part of the large complex of Pleistocene–Holocene shield
volcanoes, nested calderas, small composite cones, cinder
cones, and pit craters that are collectively referred to as
Masaya caldera (Fig. 1). This complex is part of the
Central America volcanic arc, which is characterized
during the Quaternary by predominantly basaltic volcanic
systems formed within and along the Nicaragua Depression, a NW-trending fault zone that accommodates dextral
slip resulting from oblique subduction (DeMets, 2001; La
Femina et al., 2002). Volcanoes and faults associated with
the arc and Nicaragua Depression are the dominant
structures near Masaya caldera.
This volcano complex has been the site of tremendous
Plinian basaltic eruptions during the last ∼6 ka (Williams,
1983a; van Wyk de Vries, 1993; Walker et al., 1993;
Wehrmann et al., 2006; Kutterolf et al., 2007). Kutterolf
et al. (2007) suggested that the San Antonio Tephra, with a
volume of N 14 km3 may have led to the formation of a
6 km by 11 km, northwest trending caldera (Fig. 1).
Several explosive eruptions are documented since the
formation of the caldera that might have contributed to its
subsidence, including the large-volume phreatomagmatic
deposits of the Masaya Tuff, erupted b 2 ka (Pérez and
Freundt, 2006; Kutterolf et al., 2007). These eruptions
were large volume (12–18 km3 dense rock equivalent)
and resulted in widespread tephra fallout and pyroclastic
flows (Williams, 1983a).
A prominent caldera rim was formed during these
events. The rim is up to 400 m high on the west and
northwest side of the caldera. It is ∼200 m high above
Laguna Masaya (Fig. 1) on the eastern side of the
caldera. On the south side, the caldera rim has less relief
and is partially buried by post-caldera lava flows and
tephra from post-caldera vents to the SW. The caldera
rim is subdued, or buried, on the north side of the
caldera and its exact position is inferred. The caldera
walls, where exposed, consist of thin aa-pahoehoe lava
Table 1
Summary of TEM soundings collected in Masaya caldera. Map coordinates are given in UTM Zone 16N WGS84. Depth to water table is estimated
from two separate inversion algorithms. For relative locations within the caldera see Fig. 2
Sounding #
Easting
Northing
Loop size
(m)
Land surface elevation
(masl)
Depth to water
(m) Einvrt 6
Depth to water
(m) EMVision
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
0592483
0591925
0591987
0590327
0591718
0591970
0595709
0594825
0594284
0594861
0595063
0594507
0594096
0599840
0592399
0593029
0592132
0590050
0599466
0599067
0590572
0590072
0591384
0594511
0595648
0590421
0590325
0590157
0590423
0590093
1326524
1327235
1327689
1325187
1324863
1325700
1325804
1326305
1326177
1322311
1323995
1324507
1325103
1327928
1328274
1328183
1328845
1324351
1324338
1325056
1328827
1328554
1329045
1321208
1325386
1324668
1325698
1325997
1326424
1327779
100
100
100
100
100
100
40
100
100
40
40
100
100
100
100
100
100
100
100
100
100
100
100
40
40
100
100
100
100
100
3322 80
280
278
482
569
478
162
196
228
155
176
206
228
310
285
288
305
422
390
336
311
311
322
162
120
522
459
420
358
322
51.3
141.5
172
143
218.7
243.1
46
78.8
111.9
–
48.1
70.3
88.5
164.4
175
169.2
152.1
244.8
243.6
219
171.8
176.2
165.4
7.8
5.9
294.8
183.5
221
202.3
185
59.1
141
173.1
132.1
234
255.3
46.3
88.8
122.6
–
51.2
84.1
98.8
169.5
179.2
173
158.5
241.5
235.4
216.1
177.4
179.5
172.3
18.6
21
285.4
195.1
244.6
208.7
193.1
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
221
Fig. 3. Examples of apparent resistivity curves used for preliminary interpretation of TEM soundings at Masaya caldera. (a) Most apparent resistivity
curves from TEM soundings at Masaya caldera show a highly resistive upper layer underlain at depth by one or more conductive layers (TEM
sounding 3 shown). On the southwestern side of the historically active vents, near the older Piedra Quamada lava flow, (b) the uppermost layer has a
slight conductive zone, probably produced by either alteration of the Piedra Quemada flow or the influence of the less-resistive, underlying San Judas
Formation tephra deposits (TEM sounding 18 shown). Multiple curves are shown for repeated measurements at the same stations.
flows and pyroclastic fall and flow deposits (Williams,
1983b). The total amplitude of vertical offset across
these caldera-bounding faults is not clear from geologic
outcrops or other data. Slip may have been very large if
the caldera formed predominantly by foundering during
evacuation of a large magma chamber (Williams,
1983a). Post-caldera activity included the eruption of
lava flows from caldera-bounding faults mapped by
Williams (1983b) on the south and north flanks of the
caldera (Walker et al., 1993). Thus lithologic and
hydrologic properties are likely to change abruptly
across the caldera boundary.
Volcanic activity persisted after these caldera-forming
eruptions. The floor of the caldera slopes gently to the east
(Fig. 2) and is armored by thin aa-pahoehoe basaltic lava
flows, largely erupted from a group of vents within the
caldera. These vents form a ∼5-km-long, W- to NWtrending, semi-circular group of low volcanic cones. Pit
craters at the summit of these cones, including Masaya,
Santiago, Nindiri, and San Pedro pit craters, were the
locus of historical eruptions in 1670 and 1772 that formed
10–15 km-long lava flows (Rymer et al., 1998). The
elevation of the crater rims along this chain of pit craters
varies from 500 to 635 masl, and heights of these crater
rims are ∼140–275 m above the surrounding caldera
floor (Fig. 1). Each of these pit craters is 400–1000 m in
diameter and has very steep walls, with crater depth
varying between 200–300 m.
Today, active degassing is almost entirely limited to
Santiago pit crater and a lava lake is occasionally visible in
vents through the floor of the crater. Based on time-series
of gravity observations on and about this cone complex,
222
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Fig. 4. Graphical example of analytical solution used at TEM 26 that transforms apparent resistivity values to resistivity with approximate depth
values (Meju, 1998).
Rymer et al. (1998) and Williams-Jones et al. (2003)
concluded that much of the pit-crater alignment is
underlain by a network of gas-rich, vesiculated magma
extending from approximately 200 masl (the elevation of
the floor of the Santiago pit crater) to ∼200 mbsl. This
shallow crystallizing magma is the source of persistent
degassing from Santiago crater since 1852 (Rymer et al.,
1998) and infrequent small explosions. Rymer et al.
(1998) and Williams-Jones et al. (2003) attribute these
small explosions to purely magmatic processes. They
suggest that the explosions result from blockages in the
degassing vent just below the surface. Pressure builds until
a small explosion clears the vent. There also are anecdotal
reports that these explosions occur preferentially during
the rainy season and may be associated with recharge of
the groundwater system. However, explosions have not
occurred in response to exceptional recharge events, such
as that associated with hurricane Mitch in 1998.
Although occasional explosions at Santiago pit crater
may conceivably be influenced by interaction with
meteoric water, this large and complex magmatic system
otherwise appears to be in equilibrium with the
groundwater system. On the NE flank of Masaya
volcano, within the caldera, a fault and fracture zone
hosts outflow of water vapor and CO2 at moderate
temperatures (40–70 °C), indicating that groundwater
interacts with magmatic heat, and perhaps volcanic
gases, well outside the pit crater system (Lewicki et al.,
Fig. 5. Inverse modeling results for TEM 26 using the program by Sandberg (1988). The apparent resistivity definition used is the late-stage solution
from Eq. (1). The squares are observed values and the solid line is the model fit.
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
223
Fig. 6. Comparison of the two inversion programs show consistent results with the EINVRT 6 program modeling the water table an average of 8.2%
shallower than the EM Vision models. See Table 1.
2003, 2004). Nevertheless, other than along this one
fracture zone, there is no evidence that an extensive
hydrothermal system is present within caldera outside
the active vent complex.
Data from a regional ground water flow model
developed in 1993 by the Instituto Nicarguense de
Acueductos y Alcantarillados (ENECAL) and the
Japanese International Cooperation Agency (JICA)
indicate that regional groundwater levels vary from
∼ 190 masl south of the caldera to ∼ 130 masl north of
the caldera (Fig. 1) (ENACAL and JICA, 1993).
Rainfall is high in the caldera (∼1600 mm/yr), yet no
persistent surface water flow occurs in or around it.
Regional evapotranspiration rates are high (∼ 1100 mm/
yr). No borehole data exist within the caldera itself, and
prior to this study the only indication of water levels in
the caldera was the level of Lake Masaya (Fig. 1), which
is monitored monthly by ENACAL. Lake levels during
this investigation were ∼ 119 masl. This lake level is
below the regional groundwater table (Fig. 1), suggesting that the lake is not perched, but represents the level
of the groundwater table in the caldera.
3. Mapping groundwater with the transient
electromagnetic method
et al., 2002). In recent geoelectrical studies of volcanic
systems, TEM has provided high-resolution data in the
near surface (b 1 km), and has been used to identify
hydrothermal fluid circulation and aquifer systems. The
correlation of borehole data and TEM sounding depths
for low resistivity regions on Kilauea volcano, Hawaii,
Mt. Somma-Vesuvius, Italy, Piton de la Fournaise
volcano, Réunion, and Newberry volcano, Oregon,
have shown the effectiveness of this method (Fitterman
et al., 1988; Kauahikaua, 1993; Lenat et al., 2000;
Manzella et al., 2004). Nevertheless, care is required to
assure that estimates of depth to the water table are not
affected by clay alteration in the stratigraphic section or
perched water zones. At Masaya caldera, the volcanic
geology and hydrologic conditions create a situation in
which deposits with high electrical resistivities host a
water table at depths N100 m. Porous, saturated lava
flows and scoria of the caldera floor offer an excellent
electrical contrast with the overlying, dry basaltic rocks.
Deeply exposed walls in the active vent show little
evidence of clay alteration in the edifice. Also, most of
the caldera that is close to land surface is at ambient
temperature, and Lake Masaya provides an excellent
water level for calibration of nearby TEM soundings.
4. TEM soundings
The transient electromagnetic method (McNeill,
1982, Kaufmann and Keller, 1983, Fitterman and
Stewart, 1986) has been used extensively in geophysical
groundwater exploration and, along with other electrical
resistivity methods, has been used to investigate shallow
hydrothermal systems on volcanic edifices (Sakkas
Multiple soundings of different frequencies were
made at 30 sites throughout the caldera (Table 1, Fig. 2).
TEM data were collected using a Geonics Limited
Protem 47 Digital Time Domain EM system and highfrequency receiver coil (McNeill, 1980). We used the
224
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
central-loop sounding-mode configuration, because it
has been used extensively and effectively in other
groundwater studies (Kaufmann and Keller, 1983;
Fitterman and Stewart, 1986). Data were collected at
20 logarithmically spaced time gates following transmitter turnoff. Transmitter loops were square, 40 m on a
side near Laguna Masaya and 100 m throughout the
remainder of the caldera, where the anticipated depth to
the water table was greater. Data were collected at three
different base frequencies: 30, 75, and 285 Hz, with the
receiver gain set to minimize environmental noise, yet
generate a survey signal appropriate for the given
geologic terrain without distortion to the measurement.
Output current on the Protem 47 transmitter was set from
1 Amp (A) to 3 A when using the smaller 40 m loops and
at 3 A for 100 m loops. Multiple soundings and signal
stacking were performed at each site to maximize the
signal-to-noise ratio and to evaluate reproducibility of
the sounding.
The TEM sites are situated throughout the caldera,
with an emphasis on the historically active crater region
and its flanks. Several sites were adjacent to Laguna
Masaya for verification of depth-to-water-table estimates. The remaining TEM sites are in the northern and
northeastern sections of the caldera where there is little
change in vertical relief. No soundings were made in the
southeastern and extreme western portions of the
caldera due to the difficulty of the terrain and lack of
access, but these regions are similar in elevation and
slope to the northern and eastern areas.
5. Data processing
All TEM data were transformed to apparent resistivity for preliminary interpretation, data quality assessment, and to assist in initial parameters for layered-earth
inverse models (Fig. 3a and b). The data were
transformed using the technique of Sandberg (1988),
Fig. 7. Apparent resistivity cross-sections along profiles (a) A–A’ in the eastern caldera near Laguna Masaya and (b) B–B’ near the active vent
region. Apparent resistivity vales are shown based on the Meju (1998) solution. The transition from light to medium gray represents the estimated
elevation of the water table, based on inverse modeling results (Table 1).
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
225
Fig. 8. Contour map of the estimated groundwater table at Masaya caldera from TEM measurements. The map is intended to show the general pattern
of water-table elevation, and was generated from a statistically kriged data field that, at some points, varies from the measured data by several meters.
which accounts for the finite transmitter-turnoff ramp.
For large sample times (t), and/or high resistivities (ρ),
this apparent resistivity definition asymptotes to the socalled late-stage approximation calculated by:
2=3
qlate
a ¼
a4=3 AR l5=3
202=3 p1=3 t 5=3 Z 2=3
ð1Þ
where a is the equivalent circular transmitter-loop radius,
AR is the effective area of the receiver coil, μ is the
magnetic permeability (the free-space value,
μ0 = 4π × 10− 7, was used), t is the time since transmitter
current turnoff, and Z is the mutual impedance (voltage in
the receiver coil divided by the current in the transmitter
loop).
Most apparent resistivity versus time curves from
TEM soundings at Masaya caldera show a highly resistive
upper layer underlain at depth by one or more conductive
layers (Fig. 3a). However, on the southwestern side of the
historically active vents, near the older Piedra Quamada
lava flow, the uppermost layer has a slight conductive
zone. This zone is probably produced either by alteration
of the Piedra Quamada lava flow or by the influence of the
less resistive underlying San Judas Formation tephra
deposits described by Williams (1983a) (Fig. 3b).
The TEM data were modeled using two independent
methods to estimate resistivity as a function of depth.
The first method utilizes an analytical solution that
transforms apparent resistivity values to resistivity with
approximate depth values (Meju, 1998). The pattern of
high and low resistivity layers is much more obvious in
apparent resistivity versus depth curves (Fig. 4)
calculated using the Meju (1998) solution.
The results are used to produce contoured resistivity
cross sections and to compare modeling methods. The
second method uses two commercially available programs (EM Vision from Encom and EINVRT 6 from
Geophysical Solutions) that employ non-linear, leastsquares regression algorithms to adjust layered-earth
parameters, such as layer thickness and resistivity values,
through an iterative process to minimize error (Sandberg,
1988). The EINVRT 6 program utilizes a Marquardt-type
method that results in an undamped 95% confidence level
(Hohmann and Raiche, 1988; Sandberg, 1988, 1993;
Fig. 5). The method used by EM Vision relies on the
GRENDL algorithm for 1D inversion and outputs a 68%
confidence level. All models are 1-D interpretations and
were run with the fewest number of layers that give a
reasonable fit to field data.
6. Discussion of TEM
The two inversion programs show consistent results
and comparison of the inversion models to the analytical
226
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Fig. 9. The three-dimensional groundwater flow model of Masaya caldera, showing the (a) horizontal, and (b) vertical distributions of the hydraulicparameter zones. See Table 2.
Meju solution shows similar results for most TEM
soundings. EINVRT 6 modeled the water table as being
an average of 8.2% shallower than the EM Vision
models (Fig. 6). This translates to a mean difference of
7.8 m between the two models. Due to the greater
constraint on the data by the EINVRT 6 program with its
calculated undamped 95% confidence intervals, we
believe it provides a better statistical model of the
system. The overall accuracy of the TEM method varies
across the survey area. Near the lake the lake,
comparison of TEM model results and lake levels
indicates an accuracy of within 2 m (e.g., sounding #7,
Table 1). In contrast, near the active Santiago vents the
water table is much deeper, the electrical structure is
more complex, and the accuracy is likely no better than
tens of meters (e.g., Sounding #4, Table 1).
Models of the TEM data indicate an overlying highly
resistive layer throughout the caldera that is underlain by
one or more conductive layers. The depth to the
conductive layers increases with distance from Laguna
Masaya towards the topographically high volcanic
vents. The large apparent resistivity contrast in the
subsurface is interpreted to be the interface between dry
overlying basalt and the underlying saturated zone
below the water table. This contrast is shown by the
shallow conductive layer (b118 ohm-m) detected by
TEM soundings 7 and 11 near Laguna Masaya (Fig. 2).
This conductive layer coincides with the elevation of the
nearby lake at ∼119 masl, and is assumed to be the top
of the water table (Fig. 7). Under this assumption, the
top of the conductive layer throughout the caldera is
interpreted to be the top of the saturated zone. The water
table in the caldera is nearly flat except in the higher
vent regions in the central and southern portions of the
caldera. Across the elevated vent region, the water table
or conductive zone is a subdued replica of the
topography. The resistivity cross section there (Fig. 7b)
shows a large variation in the water table between TEM
26 and TEM 4 along profile B–B/, near the active
Santiago vent. The apparently shallower water table and
lower resistivities found at TEM 4 may be due a vapordominated zone near the active vent, although the 1-D
inversion cannot fully capture the 3-D complexity of the
near-vent region.
The TEM-based water table in the caldera is very
different from the regional water levels projected from
the JICA groundwater model. In that model, the regional
ground water flow system is not affected by the presence
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Table 2
Hydraulic parameters calibrated for the groundwater model of the
caldera. For zone locations see Fig. 9
Zone
number
Hydraulic
conductivity
(m/day)
Intrinsic
permeability
(m2)
Transmissivity
(m2/s)
1
2
650
.03
7.7 × 10− 10
3.5 × 10− 14
9 × 10− 1
4 × 10− 5 to
6 × 10− 5
7 × 10− 6 to
3 × 10− 5
4 × 10− 5 to
6 × 10− 5
2 × 10− 2
3
.0054
6.4 × 10− 15
4
.023
2.7 × 10− 14
5
45
5.3 × 10− 11
of Masaya caldera (Fig. 1). Our study indicates head
values (Fig. 8) as much as 60 m lower than those
proposed from the JICA data (Fig. 1) on the western,
southern, and eastern boundaries of the caldera and
slightly lower values (10–20 m), on the northern side of
the caldera, where the caldera rim has no surface
expression. The inference of a large change in head
values across the caldera boundaries suggests that the
caldera walls are hydrologic barriers, so that the caldera
is hydrologically isolated from the surrounding region.
Masaya caldera differs from most other basaltic shield
volcanoes in having a large volume of water vapor
emissions under quiescent conditions (Burton et al.,
2000). A substantial proportion of the water vapor may
be of meteoric origin. If the caldera is hydrologically
isolated, evapotranspiration in the caldera and degassing
of meteoric water from the active vents must balance
recharge to the system. Consequently, changes in
volcanic activity, such as increased vaporization of
groundwater due to heating, should be reflected in
changes in the level of the water table. Our assessment of
the elevation of the water table at Masaya caldera is a first
step in looking at the shallow hydrologic system. A
second step is to quantify the flux of water through the
system. We use numerical simulation of groundwater
flow in the caldera to compare estimates of fluxes and
water-table elevations.
7. Groundwater model formulation
The groundwater system in Masaya caldera is highly
complex. The volcanic edifice is highly heterogeneous
and contains fault and fracture zones that have a visible
impact on water and vapor transport. High temperatures in
the active crater create a steam-dominated zone and affect
fluid density in the surrounding regions. To incorporate all
of these effects into one model would require a multiphase
code that can incorporate complex three-dimensional
227
heterogeneous terrains. The objective of this study was to
gain an understanding of the flow system at a more basic
level; such an objective can be met using a simpler,
constant-density, single-phase modeling approach. To
accomplish this, the USGS code MODFLOW2000
(Harbaugh et al., 2000) was used to construct a threedimensional flow model of the caldera.
A finite-difference grid was constructed of 10 layers,
64 rows, and 113 columns, and was aligned along the
long axis of the crater to optimize the percentage of
active grid cells (Fig. 9). Each grid cell is about 100 m in
each horizontal dimension. In the vertical dimension the
cells are about 12 m thick beneath most of caldera, but
are thicker beneath the edifice by an amount that is
proportional to the entire model thickness (Fig. 9b).
Information on lithologies and their hydraulic characteristics at depth inside the crater are nonexistent, so the
bottom of the model was arbitrarily chosen to be at sea
level (a simplification that is consistent with the rest of
this modeling scope). As MODFLOW2000 does not
consider variable saturation in the soil zone, the top of
such a model is usually placed either at the land surface
or at the approximate level of the water table. The
former option is accompanied by a variable saturation
condition in the top layer(s), whereas the latter assumes
a confined condition. The latter option was chosen for
this study to avoid the nonlinear convergence problems
that sometimes occur with cells that become dry during
modeling in MODFLOW2000. Lake Masaya was
treated as a constant head boundary, and the caldera
Table 3
Estimates of water fluxes at Masaya Caldera
Flux type
Source
of data
In
kg/s
Rainfall
Measured
Evapotranspiration Rainfall minus
from land surface recharge, and
consistent with
Maraux et al.
(1998)
Recharge
Steam discharge
plus groundwater
flow to
Laguna Masaya
665
0.5
Evaporation from
Eagleman
795
Laguna Masaya (1967)
Groundwater flow Evaporation at
265
into Laguna
lake minus
Masaya
rainfall at lake
Steam discharge
Burton et al.
400
at active vent
(2000)
In m3/year
per m2
of land
surface
In m3/
year per
m2 of
lake
1.6
1.1
1.6
2.4
0.2
0.3
0.8
228
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Fig. 10. Mean monthly water levels of Laguna Masaya for the years 1998–2003.
boundary (lateral boundary) was assigned a generalhead boundary condition. This condition allows the
leakage of water through the caldera wall to be adjusted
to create a match between water levels inside and
outside the caldera. As there is actually a great disparity
between these two sets of water levels, the leakance was
set low enough that the caldera walls act as a virtual noflow boundary. In reality, a small amount of groundwater may cross the caldera walls, especially at the eastern
end, around Laguna Masaya.
The caldera was divided into five hydraulic conductivity zones to represent the different lithologic
hydraulic conditions that exist throughout the caldera
(Fig. 9). No subsurface borehole data is available, so
these zones were chosen to represent the overall geology
as observed from the surface and the behavior of the
groundwater system near the active vent. The largest
zone (zone 1) is dominated by lava flows in the northern
and eastern caldera. The flanks of the edifice were
divided into two sections (zones 2 and 4), and the main
central edifice was defined separately (zone 3). The land
surface within the active craters was represented by drain
cells (not shown). This was done to represent the fact that
groundwater does not accumulate as a lake in the craters,
but is drawn down by evapotranspiration and thermal
vaporization. Simulation results indicated that a fifth zone
was required within the edifice to allow groundwater to
flow toward the active vent without depleting water
higher in the edifice. The parameter estimation process
in MODFLOW2000 was used to adjust the values of
hydraulic conductivity for the zones to best fit the water
levels estimated by the TEM surveys. The values that
were converged upon are listed in Table 2. Also listed are
the transmissivities, which may be a more useful result,
given the lack of data on vertical lithologic variability and
depth of the active flow system. The value for zone 1
contains the greatest uncertainty, as the very flat water
table indicates a relatively high but poorly constrained
value of hydraulic conductivity.
8. Groundwater flux estimates
Water levels alone cannot produce unique values of
hydraulic conductivity through inverse modeling of a
groundwater flow system. Flux measurements were
also required to constrain the model parameters. Flux
estimates were obtained from three sources: (1) rainfall
and evapotranspiration, (2) groundwater inflows to
Lake Masaya, and (3) steam discharge from the active
vent. The values of these fluxes are summarized in
Table 3.
Recharge was prescribed across the surface of the
caldera as a fraction of the high rainfall in the region
(1.6 m per year). Because no data exist for groundwater
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
recharge fluxes in the caldera, a value of 0.5 m/yr was
selected based in part on studies of humid volcanoes at
Kilauea (Takasaki, 1993; Ingebritsen and Scholl, 1993)
and the Cascades Range (Ingebritsen et al., 1992;
Manga, 1997). This value also was close to a value that
was calculated by balancing the groundwater budget. In
reality, recharge in the caldera is spatially variable and
likely to be a function of vegetation cover, with recharge
in vegetated areas likely less than nonvegetated areas
because of transpiration. The open areas are often bare
lava flows, with large openings at the land surface that
would allow for quick and efficient infiltration during
229
rainfall events. The evapotranspiration rate (Table 3) is
not an input for this model, but can be calculated as the
rainfall rate minus the recharge rate. For this case the
value of 1.1 m/year is similar to other values reported in
Nicaragua (e.g., Maraux et al., 1998).
Direct water flux measurements at Laguna Masaya
were not available, but some estimates could be made
using other flux components. Evaporation from the lake
surface can be estimated from climatic variables (e.g.,
Eagleman, 1967). Surface water inflow to the lake is
minor, so that a long-term steady-state balance requires
that rainfall plus groundwater inflow equal evaporation.
Fig. 11. Simulated groundwater levels (hydraulic heads) in (a) layer 1, and (b) layer 6 of the ground water model. Layer 1 is the shallowest and its head
values represent the water table. Layer 6 is deeper and includes the high permeable zone 5 (Fig. 9) that transmits water toward the active vent. The
distribution of hydraulic heads in layers 1 and 6 reflect our best estimates of the hydraulic conditions within the caldera.
230
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Thus a groundwater inflow rate could be estimated
based on the other two major fluxes. A completely
closed lake would cause build-up of dissolved salts in
the lake over time and a saline Laguna Masaya would
result. As this is not the case, there is likely some small
amount of groundwater discharge to the regional flow
system that we cannot quantify at this time. Seasonal
changes in the lake level show a steady decline during
the months of January through April (Fig. 10). Given
that precipitation during this dry season is nearly zero,
mass balance requires that this rate of lake level decline
(1.6 m/year) be nearly equivalent to the precipitation
rate, as is the case.
The final water flux component within the caldera is
that of the steam exiting at the active vent in Santiago
Crater. This flux was estimated to be approximately
400 kg/s by earlier investigations (Burton et al., 2000).
Although some of this vapor may have a magmatic
origin, its magnitude suggests a high likelihood that
most of it is of meteoric origin. Thus we used this value
to define a point discharge of groundwater in the model.
The point discharge was represented by a well at the
active crater discharging from zone 5 (Fig. 9) at a rate of
400 kg/s. This is a large fraction of the water budget,
roughly 60% of recharge within the caldera (Table 3).
The remaining 40% is assumed to discharge to Laguna
Masaya.
9. Simulation results
The final calibrated water levels and gradients are
similar to those estimated by the TEM measurements
(compare Figs. 8 and 11a). The simulated elevation of the
water table in layer 1 (Fig. 11a) reflects the calibration and
fit to the TEM data. The water-level gradient is very flat in
the outer reaches of the caldera underlain by lava flows.
Water levels beneath the flanks of the edifice begin to
increase in value, and then increase substantially beneath
the central edifice. During the model calibration process,
the water level at the active vent was not allowed to drop
below the approximate elevation of the bottom of
Santiago Crater (about 100 m amsl). Fig. 11b shows the
same hydrologic model for a deeper model layer. In order
to maintain sufficient flow toward the active vent, and
maintain discharge from the vent of approximately
400 kg/s, the hydraulic conductivity of zone 5 (Fig. 9)
had to be increased to roughly 45 m/day. As a result of the
point discharge, a cone of depression develops within
zone 5 that extends outward across the entire edifice
(Fig. 11b). On the basis of this hydraulic head distribution,
it can be inferred that much of the rainfall that recharges
the edifice may discharge out the main vent, whereas
rainfall that recharges the outer regions of the caldera may
discharge to Laguna Masaya.
10. Discussion of groundwater at Masaya
In terms of electrical structure, the caldera consists of
a highly resistive layer that is underlain by one or more
conductive layers. The top of the conductive layers is
interpreted to be the water table and is expressed as a
subdued replica of the topography in the higher vent
regions, decreasing to a level that coincides with the
elevation of Laguna Masaya in the topographically
lower areas of the caldera. There is a large (up to 60 m)
decrease in head values across the caldera boundaries,
suggesting that the caldera is hydraulically isolated from
the surrounding region. Offset across caldera-bounding
faults may be responsible for abrupt changes in
transmissivity. Alternatively, post-caldera lavas are
reported to have erupted along caldera-bounding faults
(Walker et al., 1993). Dike injection associated with
these lavas may have reduced transmissivity.
Increase in the depth to the water table near the active
vent region of Santiago crater is consistent with
vaporization of groundwater due to heating, previously
suggested to be a substantial component of the gas
emissions at Masaya caldera. The conversion of 400 kg/
s of liquid water at ambient temperatures to steam requires a very large heat input (1000 MW) that would be
originating from beneath the edifice and driving
convection of flow towards the edifice at depth.
Numerical simulations of groundwater flow in the
caldera suggest markedly different hydraulic transmissivities in different sections of the caldera. The outer
caldera, outside of the main edifice, is characterized by
basalt flows that have very high transmissivity. In order
to transmit groundwater to the active vent at a rate that is
equal to the steam discharge, a highly transmissive layer
must also exist somewhere at depth beneath the main
edifice. The groundwater model does not further
constrain the geologic nature of this transmissive layer
and it may consist of either fractured lavas or intracaldera pyroclastic units. In contrast, the rocks comprising most of the main edifice must have relatively low
transmissive properties, as low values are required to
maintain the high water table observed there. The low
values are consistent with values of young altered
volcanic rocks in Hawaii and Oregon (Ingebritsen et al.,
1992; Ingebritsen and Scholl, 1993). Alternatively, the
low values could be explained by a high water table
under perched conditions, with the calibrated transmissivities for the main edifice representing values at partial
saturation.
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
Both the estimated water table elevations and the water
budget fluxes for the caldera have substantial uncertainty.
The fluxes (except for rainfall) have uncertainties that we
estimate to be as large as 25%. Further, the multiphase
nature of the system around the active vents could not be
adequately represented in this study.
The highly connected and transmissive nature of the
subsurface at Masaya suggests that groundwater levels
may fluctuate in response to magmatic activity.
Installation of observation wells in the caldera could
facilitate a refined study of the groundwater system and
help observe and detect volcanic activity. The successful
conjunctive use of TEM and numerical simulation at
Masaya also suggests that a similar approach may be
useful at other volcanoes.
Acknowledgments
This research project at Masaya was made possible due
to financial support provided in part by the USGS Volcano
Hazards Program. We especially appreciate the assistance
of the entire staffs from the Parque Nacional Volcán
Masaya, ENECAL, and especially W. Strauch, Eduardo
Mayorga, José Manuel Traña, and Enoc Castillo from
INETER. We would also like to thank Pedro Pérez and
Félix Henrique for their invaluable help while working at
Masaya caldera, and Mark Stewart from the University of
South Florida for discussions on the hydrogeologic
system. We thank Editor Margaret Mangan, two anonymous reviewers, Steve Ingebritsen, and Shaul Hurwitz for
their help in improving this manuscript.
References
Burton, M.R., Oppenheimer, C., Horrocks, L.A., Francis, P.W., 2000.
Remote sensing of CO2 and H2O emission rates from Masaya
volcano, Nicaragua. Geology 28 (10), 915–918.
Connor, C., Hill, B., LaFemina, P., Navarro, M., Conway, M., 1996.
Soil 222Rn pulse during the initial phase of the June–August 1995
eruption of Cerro Negro, Nicaragua. Journal of Volcanology and
Geothermal Research 73, 119–127.
Delaney, P.T., 1982. Rapid intrusion of magma into wet rock:
groundwater flow due to pore pressure increases. Journal of
Geophysical Research 87, 7739–7756.
DeMets, C., 2001. A new estimate for present-day Cocos-Caribbean
plate motion: implications for slip along Central American
volcanic arc. Geophysical Research Letters 28 (21), 4043–4046.
Eagleman, J.R., 1967. Pan evaporation, potential evaporation and
actual evaporation. Journal of Applied Meteorology 6, 482–488.
ENACAL and JICA, 1993. Hydrogeological map of Managua,
Nicaragua, Instituto Nicarguense de Acueductos y Alcantarillados
and Japanese International Cooperation Agency, Managua,
Nicaragua, Scale 1:50,000.
Fitterman, D.V., Stewart, M.T., 1986. Transient electromagnetic
sounding for groundwater. Geophysics 51 (4), 995–1005.
231
Fitterman, D.V., Stanley, W.D., Bisdorf, R.J., 1988. Electrical structure
of Newberry volcano, Oregon. Journal of Geophysical Research
93 (B9), 10119–10134.
Goff, F., Janik, C.J., 2000. Geothermal systems. In: Sigurdsson, H. (Ed.),
Encyclopedia of Volcanology. Academic Press, pp. 817–855.
Harbaugh, A.W., Banta, E.R., Hill, M.C., McDonald, M.G., 2000.
MODFLOW-2000, the U. S. Geological Survey modular groundwater model-user guide to modularization concepts and the
ground-water flow process. U. S. Geological Survey Open-File
Report, vol. 00-92.
Hayba, D.O., Ingebritsen, S.E., 1997. Multiphase groundwater flow
near cooling plutons. Journal of Geophysical Research 102,
12235–12252.
Hohmann, G.W., Raiche, A.P., 1988. Inversion of controlled-source
electromagnetic data. In: Nabighian, M.N. (Ed.), Electromagnetic
Methods in Applied Geophysics, Volume 1: Investigations in
Geophysics, vol. 3. Society of Exploration Geophysicists, Tulsa,
OK, pp. 469–503.
Hurwitz, S., Kipp, K.L., Ingebritsen, S.E., Reid, M.E., 2003. Groundwater flow, heat transport, and water table position within volcanic
edifices: implications for volcanic processes in the Cascade Range.
Journal of Geophysical Research 108. doi:10.1029/2003JB002565.
Ingebritsen, S.E., Scholl, M.A., 1993. The hydrogeology of Kilauea
Volcano. Geothermics 22, 255–270.
Ingebritsen, S.E., Sherrod, D.R., Mariner, R.H., 1992. Rates and
patterns of groundwater flow in the Cascade Range volcanic arc,
and the effect on subsurface temperatures. Journal of Geophysical
Research 97, 4599–4627.
Ingebritsen, S.E., Sanford, W.E., Neuzil, C.E., 2006. Groundwater in
Geologic Processes. Cambridge University Press, Cambridge.
Kauahikaua, J., 1993. Geophysical characteristics of the hydrothermal
systems of Kilauea volcano, Hawaii. Geothermics 22 (4), 271–299.
Kaufmann, A.A., Keller, G.V., 1983. Frequency and Transient
Soundings. Elsevier, Amsterdam.
Kutterolf, S., Freundt, A., Pérez, W., Wehrman, H., Schminke, H.-U.,
2007. Late Pleistocene to Holocene temporal succession and
magnitudes of highly-explosive volcanic eruptions in west-central
Nicaragua. Journal of Volcanology and Geothermal Research 163,
55–82.
La Femina, P.C., Dixon, T.H., Strauch, W., 2002. Bookshelf faulting in
Nicaragua. Geology 30 (8), 751–754.
Lénat, J.F., Fitterman, D., Jackson, D.B., Labazuy, P., 2000.
Geoelectrical structure of the central zone of Piton de la Fournaise
volcano (Réunion). Bulletin of Volcanology 62, 75–89.
Lewicki, J., Connor, C., St-Armand, K., Stix, J., Spinner, W., 2003.
Self-potential, soil CO2 flux, and temperature on Masaya volcano,
Nicaragua. Geophysical Research Letters 30 (15), 1817–1821.
Lewicki, J.L., Hilley, G.E., Connor, C.B., 2004. The scaling
relationship between self-potential and fluid flow on Masaya
volcano, Nicaragua. In: Wanty, R.B., Seal II, R.R. (Eds.), Water–
Rock Interaction. Taylor and Francis Group, London. ISBN: 90
5809 6416, pp. 153–156.
Manga, M., 1997. A model for discharge in spring-dominated streams
and implications for the transmissivity and recharge of Quaternary
volcanics in the Oregan Cascades. Water Resources Research 33,
1813–1822.
Manzella, A., Volpi, G., Zaja, A., Meju, M., 2004. Combined TEMMT investigations of shallow-depth resistivity structure of Mt
Somma-Vesuvius. Journal of Volcanology and Geothermal
Research 131, 19–32.
Maraux, F., Lafolle, F., Bruckler, L., 1998. Comparison between
mechanistic and functional models for estimating soil water
232
R.E. MacNeil et al. / Journal of Volcanology and Geothermal Research 166 (2007) 217–232
balance. Deterministic and stochastic approaches. Agricultural
Water Management 38 (1), 1–20.
McNeill, J.D., 1980. Applications of Transient Electromagnetic
Techniques, Technical Note TN-7. Geonics Ltd, Mississauga
Canada. 15 pp.
McNeill, J.D., 1982. Use of electromagnetic methods for groundwater
studies. In: Ward, S.H. (Ed.), Geotechnical and Environmental
Geophysics, Volume II: Environmental and Groundwater, Investigations in Geophysics No. 5. Society of Exploration Geophysicists, Tulsa, OK, pp. 191–218.
Meju, M.A., 1998. A simple method of transient electromagnetic data
analysis. Geophysics 63 (2), 405–410.
Morrissey, M., Zimanowski, B., Wohletz, K., Buettner, R., 2000.
Phreatomagmatic fragmentation. Encyclopedia of Volcanology.
Academic Press, pp. 431–445.
Newhall, C.G., Albano, S.E., Matsumoto, N., Sandoval, T., 2001.
Roles of groundwater in volcanic unrest. Journal of the Geological
Society of the Philippines 56, 69–84.
Pérez, W., Freundt, A., 2006. The youngest highly explosive basaltic
eruptions from Masaya Caldera (Nicaragua): stratigraphy and hazard
assessment. In: Rose, W.I., Bluth, G.J.S., Carr, M.J., Ewert, J., Patino,
L.C., Vallance, J.W. (Eds.), Volcanic Hazards in Central America.
Geological Society of America Special Paper, 412, pp. 189–207.
Rymer, H., van Wyk de Vries, B., Stix, J., William-Jones, G., 1998. Pit
crater structure and processes governing persistent activity at
Masaya volcano, Nicaragua. Bulletin of Volcanology 59, 345–355.
Sandberg, S.K., 1988. Inverse Modeling Software for Resistivity,
Induced Polarization (IP), and Transient Electromagnetic (TEM,
TDEM) Soundings. Geophysical Solutions, Inc., Albuquerque,
New Mexico.
Sandberg, S.K., 1993. Examples of resolution improvement in
geoelectrical soundings applied to groundwater investigations.
Geophysical Prospecting 41, 207–227.
Sakkas, V., Meju, M.A., Khan, M.A., Haak, V., Simpson, F., 2002.
Magnetotelluric images of the crustal structure of Chyulu Hills
volcanic field, Kenya. Tectonophysics 346, 169–185.
Shibata, T., Akita, F., 2001. Precursory changes in well water level
prior to the March 2000 eruption of Usu volcano, Japan.
Geophysical Research Letters 28 (9), 1799–1802.
Sparks, S., 2003. Forecasting volcanic eruptions. Earth and Planetary
Science Letters 210, 1–15.
Takasaki, K.J., 1993. Ground water in Kilauea Volcano and adjacent
areas of Mauna Loa, Island of Hawaii. U. S. Geological Survey
Open-File Report, vol. 93-82.
Van Wyk de Vries, B., 1993, Tectonics and Magma Evolution of
Nicaragua Volcanic Systems, Ph.D. Thesis, Open University,
Milton Keynes, UK (unpublished).
Walker, J.A., Williams, S.N., Kalamarides, R.I., Feigenson, M.D., 1993.
Shallow open-system evolution of basaltic magma beneath a
subduction zone volcano: the Masaya Caldera Complex, Nicaragua.
Journal of Volcanology and Geothermal Research 56, 379–400.
Wehrmann, H., Bonadonna, C., Freundt, A., Houghton, B.F., Kutterolf,
S., 2006. Fontana Tephra: a basaltic Plinian eruption in Nicaragua.
In: Rose, W.I., Bluth, G.J.S., Carr, M.J., Ewert, J., Patino, L.C.,
Vallance, J.W. (Eds.), Volcanic Hazards in Central America.
Geological Society of America Special Paper, 412, pp. 209–223.
William-Jones, G., Rymer, H., Rothery, D.A., 2003. Gravity changes and
passive SO2 degassing at the Masaya caldera complex, Nicaragua.
Journal of Volcanology and Geothermal Research 123, 137–160.
Williams, S.N., 1983a. Plinian airfall deposits of basaltic composition.
Geology 11, 211–214.
Williams, S.N., 1983b. Geology and eruptive mechanisms of Masaya
Caldera Complex, Nicaragua, PhD thesis, Dartmouth College,
Hanover, NH, (unpubl.)